DESCRIPTION: Ensure that a MERGE command on a distributed table with a
`WHEN NOT MATCHED BY SOURCE` clause runs against all shards of the
distributed table.
The Postgres MERGE command updates a table using a table or a query as a
data source. It provides three ways to match the target table with the
source: `WHEN MATCHED` means that there is a row in both the target and
source; `WHEN NOT MATCHED` means that there is a row in the source that
has no match (is not present) in the target; and, as of PG17, `WHEN NOT
MATCHED BY SOURCE` means that there is a row in the target that has no
match in the source.
In Citus, when a MERGE command updates a distributed table using a
local/reference table or a distributed query as source, that source is
repartitioned, and for each repartitioned shard that has data (i.e. 1 or
more rows) the MERGE is run against the corresponding distributed table
shard. Suppose the distributed table has 32 shards, and the source
repartitions into 4 shards that have data, with the remaining 28 shards
being empty; then the MERGE command is performed on the 4 corresponding
shards of the distributed table. However, the semantics of `WHEN NOT
MATCHED BY SOURCE` are that the specified action must be performed on
the target for each row in the target that is not in the source; so if
the source is empty, all target rows should be updated. To see this,
consider the following MERGE command:
```
MERGE INTO target AS t
USING source AS s ON t.id = s.id
WHEN NOT MATCHED BY SOURCE THEN UPDATE t SET t.col1 = 100
```
If the source has zero rows then every row in the target is updated s.t.
its col1 value is 100. Currently in Citus a MERGE on a distributed table
with a local/reference table or a distributed query as source ignores
shards of the distributed table when the corresponding shard of the
repartitioned source has zero rows. However, if the MERGE command
specifies a `WHEN NOT MATCHED BY SOURCE` clause, then the MERGE should
be performed on all shards of the distributed table, to ensure that the
specified action is performed on the target for each row in the target
that is not in the source. This PR enhances Citus MERGE execution so
that when a repartitioned source shard has zero rows, and the MERGE
command specifies a `WHEN NOT MATCHED BY SOURCE` clause, the MERGE is
performed against the corresponding shard of the distributed table using
an empty (zero row) relation as source, by generating a query of the
form:
```
MERGE INTO target_shard_0002 AS t
USING (SELECT id FROM (VALUES (NULL) ) source_0002(id) WHERE FALSE) AS s ON t.id = s.id
WHEN NOT MATCHED BY SOURCE THEN UPDATE t set t.col1 = 100
```
This works because each row in the target shard will be updated, and
`WHEN MATCHED` and `WHEN NOT MATCHED`, if specified, will be no-ops
because the source has zero rows.
To implement this when the source is a local or reference table involves
teaching function `ExcuteSourceAtCoordAndRedistribution()` in
`merge_executor.c` to not prune tasks when the query has `WHEN NOT
MATCHED BY SOURCE` but to instead replace the task's query to one that
uses an empty relation as source. And when the source is a distributed
query, function
`ExecuteMergeSourcePlanIntoColocatedIntermediateResults()` (also in
`merge_executor.c`) instead of skipping empty tasks now generates a
query that uses an empty relation as source for the corresponding target
shard of the distributed table, but again only when the query has `WHEN
NOT MATCHED BY SOURCE`. A new function `BuildEmptyResultQuery()` is
added to `recursive_planning.c` and it is used by both the
aforementioned functions in `merge_executor.c` to build an empty
relation to use as the source. It applies the appropriate type to each
column of the empty relation so the join with the target makes sense to
the query compiler.
In preparation of sorting and grouping all includes we wanted to move
this file to the toplevel includes for good grouping/sorting.
(cherry picked from commit 0dac63afc0)
PG16 compatibility - Part 6
Check out part 1 42d956888d
part 2 0d503dd5ac
part 3 907d72e60d
part 4 7c6b4ce103
part 5 6056cb2c29
This commit is in the series of PG16 compatibility commits.
It handles the Permission Info changes in PG16. See below:
The main issue lies in the following entries of PlannedStmt: {
rtable
permInfos
}
Each rtable has an int perminfoindex, and its actual permission info is
obtained through the following:
permInfos[perminfoindex]
We had crashes because perminfoindexes were not updated in the finalized
planned statement after distributed planner hook.
So, basically, everywhere we set a query's or planned statement's rtable
entry, we need to set the rteperminfos/permInfos accordingly.
Relevant PG commits:
a61b1f7482
a61b1f74823c9c4f79c95226a461f1e7a367764b
b803b7d132
b803b7d132e3505ab77c29acf91f3d1caa298f95
More PG16 compatibility commits are coming soon ...
1) For distributed tables that are not colocated.
2) When joining on a non-distribution column for colocated tables.
3) When merging into a distributed table using reference or citus-local tables as the data source.
This is accomplished primarily through the implementation of the following two strategies.
Repartition: Plan the source query independently,
execute the results into intermediate files, and repartition the files to
co-locate them with the merge-target table. Subsequently, compile a final
merge query on the target table using the intermediate results as the data
source.
Pull-to-coordinator: Execute the plan that requires evaluation at the coordinator,
run the query on the coordinator, and redistribute the resulting rows to ensure
colocation with the target shards. Direct the MERGE SQL operation to the worker
nodes' target shards, using the intermediate files colocated with the data as the
data source.
Baseinfo also has pushed down filters etc, so it makes more sense to use
BaseRestrictInfo to determine what columns have constant equality
filters.
Also RteIdentity is used for removing conversion candidates instead of
rteIndex.
AllDataLocallyAccessible and ContainsLocalTableSubqueryJoin are removed.
We can possibly remove ModifiesLocalTableWithRemoteCitusLocalTable as
well. Though this removal has a side effect that now when all the data
is locally available, we could still wrap a relation into a subquery, I
guess that should be resolved in the router planner itself.
Add more tests
We should not recursively plan an already routable plannable query. An
example of this is (SELECT * FROM local JOIN (SELECT * FROM dist) d1
USING(a));
So we let the recursive planner do all of its work and at the end we
convert the final query to to handle unsupported joins. While doing each
conversion, we check if it is router plannable, if so we stop.
Only consider range table entries that are in jointree
If a range table is not in jointree then there is no point in
considering that because we are trying to convert range table entries to
subqueries for join use case.
Check equality in quals
We want to recursively plan distributed tables only if they have an
equality filter on a unique column. So '>' and '<' operators will not
trigger recursive planning of distributed tables in local-distributed
table joins.
Recursively plan distributed table only if the filter is constant
If the filter is not a constant then the join might return multiple rows
and there is a chance that the distributed table will return huge data.
Hence if the filter is not constant we choose to recursively plan the
local table.
When doing local-distributed table joins we convert one of them to
subquery. The current policy is that we convert distributed tables to
subquery if it has a unique index on a column that has unique
index(primary key also has a unique index).
The logical planner cannot handle joins between local and distributed table.
Instead, we can recursively plan one side of the join and let the logical
planner handle the rest.
Our algorithm is a little smart, trying not to recursively plan distributed
tables, but favors local tables.
Before this commit, Citus supported INSERT...SELECT queries with
ON CONFLICT or RETURNING clauses only for pushdownable ones, since
queries supported via coordinator were utilizing COPY infrastructure
of PG to send selected tuples to the target worker nodes.
After this PR, INSERT...SELECT queries with ON CONFLICT or RETURNING
clauses will be performed in two phases via coordinator. In the first
phase selected tuples will be saved to the intermediate table which
is colocated with target table of the INSERT...SELECT query. Note that,
a utility function to save results to the colocated intermediate result
also implemented as a part of this commit. In the second phase, INSERT..
SELECT query is directly run on the worker node using the intermediate
table as the source table.
With this commit, Citus recursively plans subqueries that
are not safe to pushdown, in other words, requires a merge
step.
The algorithm is simple: Recursively traverse the query from bottom
up (i.e., bottom meaning the leaf queries). On each level, check
whether the query is safe to pushdown (or a single repartition
subquery). If the answer is yes, do not touch that subquery. If the
answer is no, plan the subquery seperately (i.e., create a subPlan
for it) and replace the subquery with a call to
`read_intermediate_results(planId, subPlanId)`. During the the
execution, run the subPlans first, and make them avaliable to the
next query executions.
Some of the queries hat this change allows us:
* Subqueries with LIMIT
* Subqueries with GROUP BY/DISTINCT on non-partition keys
* Subqueries involving re-partition joins, router queries
* Mixed usage of subqueries and CTEs (i.e., use CTEs in
subqueries as well). Nested subqueries as long as we
support the subquery inside the nested subquery.
* Subqueries with local tables (i.e., those subqueries
has the limitation that they have to be leaf subqueries)
* VIEWs on the distributed tables just works (i.e., the
limitations mentioned below still applies to views)
Some of the queries that is still NOT supported:
* Corrolated subqueries that are not safe to pushdown
* Window function on non-partition keys
* Recursively planned subqueries or CTEs on the outer
side of an outer join
* Only recursively planned subqueries and CTEs in the FROM
(i.e., not any distributed tables in the FROM) and subqueries
in WHERE clause
* Subquery joins that are not on the partition columns (i.e., each
subquery is individually joined on partition keys but not the upper
level subquery.)
* Any limitation that logical planner applies such as aggregate
distincts (except for count) when GROUP BY is on non-partition key,
or array_agg with ORDER BY
Colm
Jelte Fennema-Nio
Nils Dijk
Naisila Puka
Teja Mupparti
SaitTalhaNisanci
Marco Slot
Onder Kalaci
Hadi Moshayedi
Philip Dubé
Onder Kalaci